qst6105.tmp Productivity Growth and Input Mix Changes in Food Processing Adesoji O. Adelaja To examine productivity growth in New Jersey’s food-processing sector, this study conducts a joint analysis of total and partial factor productivity indexes. Results indicate growing material intensity, declining labor and capital intensities, and relatively slow material productivity growth. However, due to the high cost share of material inputs, material productivity growth contributed more to total factor productivity growth than did growth in the productivity of any other input. In fact, almost half of the growth in overall productivity is attributed to material productivity growth. Results also suggest that the 1973 decline in total factor productivity was characterized by greater decline in material productivityy than in the productivities of labor and capital. Changes in labor and total factor productivity have been the focus of several studies on the U.S, food- processing sector. Results of these studies gener- ally indicate that while labor and total factor pro- ductivity increased over time, both declined in 1973 and in the period immediately following. The temporary declines in these productivity indexes have been attributed to the supply shock resulting from the 1973 energy crisis (Lee; Jorgenson, Gol- lop, and Fraumeni; Heien). A major limitation of these studies, however, is that they ignored (1) the behavior of productivity indexes for nonlabor in- puts and (2) the effects of energy prices on such productivity indexes. In most manufacturing industries, labor inten- sity is high. In conducting productivityy analysis in these industries, it makes sense to focus on labor productivity. In food processing, however, mate- rial inputs account for over 6070 of production cost (Adelaja 1992). It does not make intuitive sense for productivity studies to focus on labor produc- tivity because material productivity growth is probably more relevant than labor productivity growth, and gains in material efficiency are likely to have greater effect on total factor productivity growth than do gains in labor efficiency. Knowledge of the behavior of productivity in- dexes for nonlabor inputs and of the contributions Adesoji 0. Adelaja is an assistant professor in the Department of Agricultural Economics and Marketing, Cook College, Rutgers Univer- sity. The author wishes to thank Susan Howard for her word-processing assistance. New Jersey Agricultural Experiment Station Publication no. 0-02134-5-91, supported by a grant from the New Jersey Department of Agriculture and state experiment station funds appropriated under the Hatch and McIrNire-StennisActs. of these indexes to total factor productivity growth is useful to economists in understanding the nature and sources of productivity growth in food pro- cessing. Information on the impact of energy price shocks on productivity indexes for nonlabor inputs could also contribute to knowledge about the role of energy inputs in the food-processing sector. Ag- ricultural economists should particularly be inter- ested in productivity indexes for material inputs because 70% of materials used in U.S. food pro- cessing are farm products (Adelaja 1992). That is, material productivity indexes should reflect the dy- namics of the efficiency of use of farm products in food processing. For example, the contributions of efficiency gains in material use to total factor pro- ductivity gains should be of interest to agricultural economists. Using the state of New Jersey as a case study, this paper estimates and analyzes changes in total factor productivity as well as productivity indexes for four classes of food-processing inputs: produc- tion labor, nonproduction labor, capital, and ma- terials. For each year in the 1964-84 period, these productivity indexes are derived for the aggregate food-processing sector (SIC 20) and for each sub- sector (three-digit SIC categories). 1 To facilitate decomposition of growth in total factor productiv- ityy indexes into growth in partial factor productiv- 1 Growth in New Jemey’s food-processing sector has been slower than in ttre U.S. New Jersey’s shares of U.S. food-processing employment, value of shipments, and value-added fell from 4.270, 4.0%, and 4.5% in the early 1970s to 3.8%, 2.9%, and 3.7%, respectively, by 1984. New Jersey’s share of U.S. population remained constant at 3 .4% during this period (Annual Survey of Manufacturers,U.S. Department of Com- merce). 22 April 1992 N.IARE ity indexes, the theoretical relationship between the two is derived. This decomposition allows one to observe the extent to which efficiency gains in the use of a specific input contribute to total factor productivity growth. By further focusing on pro- ductivity changes in the 1972–73 period, the im- mediate impacts of the energy crisis are further examined. Results illuminate the structure of pro- ductivity growth and some of the implications of such growth. The plan for the rest of this paper is as follows. The theoretical relationships between total and par- tial factor productivity indexes are derived in the following section. The empirical model used in productivity analysis and decomposition appears next followed by a discussion of the data and the empirical results. The final section presents con- cluding remarks. Total and Partial Factor Productivity Indexes Denote the total factor productivity index for the tth period as TFPt and the partial factor produc- tivity index for the ith input in the t th period as PFPi,. TFPt indexes reflect overall efficiency gains (in the utilization of all inputs). On the other hand, PFPi, indexes reflect not only efficiency gains in the utilization of specific inputs, but also changes in input mix due to technological biases and input substitution.2 Because TFPt and PFPi, are related under certain conditions, one can decompose changes in overall efficiency into changes in effi- ciencies of each input. This decomposition proce- dure is outlined in the rest of this section. Assume the existence of a production function relating inputs to output (Q). Further assume com- petitive input markets (CP) and constant returns to scale (CRS). Following the convention in the An- nual Survey of Manufacturers (ASM) and Census of Manufacturers (CM), also assume four catego- ries of inputs: (1) production labor (L), nonproduc- tion labor (R), material inputs (M), and capital (K).3 Denote the quantities used of these inputs in time period tby XL,,XRI,X~,, and X~, so that X~is a vector of input quantities (Xi) in period t.For the tth period, denote the price of the ith input as Pi, (e.g., price of material inputs is P~j and out- 2 TFP,is the ratio of output (Q,) to the quantity of aggregate input in the tth year, It shows changes in aggregate input when output is held constant, PFPi,is the ratio of output (Q,) to the quantity of the ithinput (X,) in the fth year. It shows changes in the input’s quantity when output is held constant, 3 Materials include inputs that are completely exhausted in produc- tion. Nonpruduction labor includes management and service-type work- ers. put price as P.Q,..The production function can be specified irnphcltly as (1) Q, = FJX,7J In (1), Tt is the value of a trend variable in the tth period. It is therefore a proxy for technology. Following Evenson, Landau, and Ballou, obtain where Fit = aQJdXi, equals the marginal product of the ith input in the tth period. Profit maximi- zation implies that Fi, = Pi~P~,. Hence, n Pi, dXj, (3) ~.dTr=~~-~ . dTt + FTC“ dTt, t i=l ‘ and Si, is cost share of the ith input in the tth period. Under CRS, P~,Q, = ~~=, Pi,xi,, and z:= I Si,= 1. TFP growth rate (T~P,) is — FT, ‘5) ‘Fp’ = z “ ‘T: n NnQf dlllxi,.— . aT, ‘ dTt-~si,~”dTt i=1 = of - ~ Si$i,. i=1 Productivity growth rate of \he ith input (P~Pi) is dln(QJXij (6) PFPi, = aTl . dTt alnQt i)lnxi, =—. dTt–~”dTl aT, / The simple unweighed average of the PFPi, growth rates (A=PJ is Adelaja Productivity Growth and Input Mix Changes in Food Processing 23 From (5) and (7), note that n According to (8), APFPt k output growth rate minus simple average growth rate of inputs, while T~P, k output growth rate minus weighted aver- age growth rate of inputs. Derive the following from (8): n n i= 1 n Hence, T~Pt k the weighted average of P~Pi, val- ues, while #@Pt k the simple average. The dif- ference between A~Pt and T~Pt k defined as fol- lows: (11) The contribution of growth in the productivity of the ith input to total factor productivity growth is obtained from (9) as S#~Pi,. The proportion of total factor productivity growth that is due to growth in productivity of the ith input (Ci) is there- fore obtained as (12) SiP=i, Ci=— T~P$ No& that ~., Ci = 1 under~RS, If (Q~X~) > (Q/X~), then ~~ > ~~ and X~/X~ >0. Hence, changes in input ratios are reflected by differences in partial factor productivity growth rates. Partial factor productivity indexes can be used to charac- terize changes in input intensitv via intensitv mea- sures (W), ‘defined ‘as follows:- (13) Wi,= In (13), IVi is the change in the intensity of input i in time pe~od t and IUI is the absolute value of U. If Wi, > 0 (IVi, < O), production becomes more (less) input i intensive over time. Note that Z?= ~ Ivi,= o. Empirical Model The analysis in the previous section is in continu- ous time. However, it is difficult to calculate pro- ductivity indexes from continuous time-series data since both Si and Xi change between periods. To solve this problem, Jorgenson, Gollop, and Frau- meni recommend discrete approximation via the logarithmic indexing method (LIM), LIM is con- sistent with the translog production-function spec- ification of the implicit production function in equation (1). Following Jorgenson, Gollop, and Fraumeni, define T~Pt as the average growth rate of TFP between two discrete points in time, say time pe- riod (t) and (t – 1). That is, T~P, k approximated by TFPf = Y2[TFP, + TFP,. ~]. Further define S* as the average factor shares of the ith input b&ween two discrete points in time. That is, St = %[Si, + Si,_,]. Considering that TFPf k also the difference between successive logarithms of output minus the weighted average of the difference be- tween successive logarithms of inputs with the 24 April 1992 weights being the S~s (Christensen, Cummings, and Jorgensen), it can be obtained as (14) TF~ = lnQ, - lnQ1.l - ~s: [lnx,, - lnx ]1,., i=1 . QT - i s,~;, i=1 where Q~ = lnQt – lnQ,_, and X; = lnxi, – lnXi,_,. Note also that (15) @ = TFP? + ~ S,$Y$. i=1 Equations (14) and (15) show the traditional TFP decomposition relationship. CRS and CP are imposed on the translog pro- duction function via the constraint that ~~~ ~Si = 1. This constraint allows one to define TFP growth rate as the weighed average of partial factor pro- ductivity growth rates (see equation 9). These as- sumptions are not required to use LIM. However, they are required to use the TFP decomposition relationship in equation (9) since the relationship is based on the premise of CRS and CP. CRS and CP imply that the constraint 27=, Si = 1 must be imposed in using the LIM procedure. This is equivalent to the constraint that 27= ~ S? = 1. Imposition of the constraint is further discussed in the data section. P~Pi, is approximated by PFP~ = [lnQ, – lnQ,_ ~] – [lnXir – lnXi,_,] = Q: – X~. Hence from (15), (16) PFP~ = ~ – X; n–1 NJARE n–1 (18) TFF# = PFP; - ~ s~x$- X2],“ i=l Using (14) and (17), TFP~ and PFP~ can be cal- culated from time-series data on real quantities of outputs and inputs, and cost shares of inputs. In- dexes of TFPt and PFPi{ can be further con- structed. The relationships m (5) through (13) also armly to the TFP and PFP indexes obtained via d~~c~eteapproximations posed in using the LIM. Data and Calculations if CRS and CP are im- CM and ASM publish annual New Jersey data on value of shipments (VS), expenditures on materials (ME), hours of production labor employment (L.H), wages paid to production labor (Z@, wages paid to all labor (LRE), number of workers (LRN), and number of production workers (LN) for the food-processing sector (SIC 20) and each three- digit SIC category except for fats and oils (SIC 207). The data is consistently available for the years 1964 through 1984. The quantity index for the production labor in- put (XL) is obtained as the index of LH. Total number of nonproduction workers (RN) is calcu- lated as LRN – LN. The index of the nonproduc- tion labor input (X~) is obtained as the imputed hours of employment of nonproduction workers (RH), which is obtained by assuming that each nonproduction worker works 40 hours per week and 52 weeks per year. X~ is obtained by dividing ME by the producer price indexes for materials and components obtained from producer price indexes (U.S. Department of Labor, Bureau of Labor Sta- tistics). This data source also provides data on pro- ducer price indexes for all food products (SIC 20) and for each three-digit SIC category of food prod- ucts. These are used as deflators for VS to obtain implicit output quantity indexes (Qt). The annual cost of capital and the index of capital input in real terms (X~) are obtained from Adelaja (1988, 1992). The base yedr for all indexes is 1964 (1964 = 100). Wages paid to nonproduction workers (ZW) are calculated as LRE – LE. Total cost of production (TC) is calculated as KE + RE + ME + LE. Input shares (Si) are calculated as KE, RE, ME, or LE divided by TC so that Z:= ~Si = 1, as required Adelaja Productivi~ GrowthartdInput Mix Changes in Food Processing 25 Table 1. Estimated Productivity Indexes for New Jersey’s Aggregate Food-Processing Sector (SIC 20), 1964-S4 Total Production Nonproduction Material Capital Factor Labor Labor Input Input Year Productivity Productivity Productivity Productivity Productivity 1964 100 100 100 100 100 1965 98 96 105 98 106 1966 96 92 102 95 109 1967 102 102 111 100 118 1968 102 103 112 101 119 1969 102 102 107 100 122 1970 101 108 120 97 134 1971 102 110 121 98 137 1972 102 112 127 97 140 1973 87 101 117 81 128 1974 93 104 119 87 136 1975 99 113 126 93 150 1976 108 131 140 101 160 1977 105 137 151 96 172 1978 107 137 150 97 174 1979 111 135 153 102 177 1980 117 135 156 109 1so 1981 124 140 161 117 187 1982 125 144 160 117 190 1983 126 144 160 118 183 1984 128 144 158 121 182 Percent Growth: 1964-84 28 44 58 21 82 1972–73 – 15 – 10 –8 – 16 –9 under the CRS and CP assumptions.4 Values of (PFP~) are used to obtain indexes of partial factor productivity (PFPi), while those of TFP* are used to obtain indexes of TFP. Validity of the translog production-function specification is assumed. Empirical Results TFP, and PFPj, indexes derived for the aggregate sector appear m Table 1.5Percentage growths of these indexes for the 1964-84 and 1972–73 peri- ods are also reported in Table 1. For the same periods, percentage growths in TFP, and PFPi, for 4 Constraints implied by CRS and CP impose strong restrictions on the characterization of food-processing technology, but they allow definition of input shares as output elasticity and definition of TFP growth rate as tbe weighted-average growth rates of PFP,. Evidence of market power and price-setting behavior in food processing appears in Azzam and Pagoulatos, Schrueter, Schroeter and Azzam, and Connor, Rogers, Mar- ion, and Mueller. Pratten, amongst others, also provides evidence of non-constant returns to scale in food processing. These suggest that TFP, and PFP,,measures derived in this analysis may be biased. For example, TFP, would be biased downwards if production is character- ized by increasing returns to scale and upwards if characterized by de- creasing returns to scale. 5 Productivity indexes generated from implicit quantity indexes are sensitive to price variation and the choice of price deflator. the subsectors appear in Table 2. Intensity values (W) for the sector and subsectors appear in Ta- ble 3. The Aggregate Sector, 1964-84 Table 1 indicates that all indicators of food- processing efficiency (all productivity measures) in New Jersey experienced secular growth during the 1964-84 period. The 28% growth in TFP in the aggregate sector is tantamount to a 21 ?40 material productivity growth, 44% production labor pro- ductivity growth, 58% nonproduction labor pro- ductivity growth, and 82% capital productivity growth. Obviously, labor productivity growth alone does not provide a full picture of productiv- ity growth in food processing. These results indi- cate that economists need to examine other partial productivity indexes to fully understand productiv- ity growth. Material productivity growth was relatively slow during the 1964-84 period (2170, compared to 4470, 58Y0, and 82% for other inputs). This suggests greater constraints in increasing the pro- ductivity of materials vis-h-vis other inputs. This phenomenon can be attibuted to the strong com- plementarily between material inputs and output, 26 April 1992 NJARE Table 2. Estimated Percentage Growth in Total and Partial Factor Productivity Indexes for the Subsectors Total Production Nonproduction Material Capital SIC Factor Labor Labor Input Input Codea Productivity Productivity Productivity Productivity Productivity 1964-s4 201 10 – 21 – 13 17 11 202 24 79 118 8 74 203 30 8 23 24 63 204 3 – 16 81 2 82 205 45 56 89 34 90 206 34 96 111 16 142 208 24 89 98 –4 130 209 22 5 1 21 77 1972-73 201 – 14 –6 –9 – 15 –8 202 –9 -6 – 18 –8 – 15 203 -9 –5 – 17 –9 – 16 204 –21 19 -23 -21 – 14 205 – 14 –11 -11 – 16 – 14 206 – 26 – 12 – 14 – 32 –4 208 – 18 – 13 –1 – 25 –11 209 – 19 –21 –1 -20 –1 Whe SIC categories areas follows: 201, meat products; 202, dairy products; 203, preserved fruit and vegetable products; 204, grain mill products; 205, bakery products; 206, sugar and confectionery products; 208, beverage products; and 209, miscellaneous and limited short-run substitution of other inputs for materials (Adelaja 1992). Food processors seem to face less constraints in increasing labor and capital productivity. Table 3. Estimated Input Intensity Values SIC Intensity Measures Codea IV, IvR IvM Iv. 1964-84 20 201 202 203 204 205 206 208 209 1972-73 20 201 202 203 204 205 206 208 209 0.14 9.50 –0.131 0.73 1.43 0.16 –0.05 –0.14 0.81 –0.09 – 0.40 –0.50 –0.58 0.00 0.38 –0.25 0.00 0.91 –0.14 5.50 -0.69 0.23 –1.19 –0.33 –0.22 –0.26 0.96 –0.27 –0.10 0.50 0.42 0.21 –2.38 –0.13 -0.92 -0.91 0.59 – 9.50 0.89 0.20 0,95 0.49 0.82 1.05 0.19 0.45 0.50 –0.33 –0.25 0.11 1.00 1.00 0.92 0.82 –0.61 –9.50 –0.06 -1.10 – 1.22 –0.34 – 0.56 –0.67 -1.96 -0.18 –0.20 0.25 0,33 –0.26 0.75 –0.75 –0.15 –0.91 *The SIC categories are as follows: 20, total for all food prod- ucts; 201, meat products; 202, dairy products 203, preserved fruit and vegetable products; 204, grain mill products; 205, bakery products; 206, sugar and confectionery products; 208, beverage products; and 209, miscellaneous products. In spite of limited material productivity growth, material productivity’s contribution to total factor productivity growth should not be downplayed be- cause of materials’ high cost share. Material pro- ductivity’s true contribution to total factor produc- tivity growth is the product of the average material factor share (.60) and total material productivity growth (21 %), divided by total factor productivity growth (28%). Hence, material productivity growth alone contributed 45% of the 28% growth in total factor productivity (12.6% TFP growth). This significant contribution to total factor produc- tivity growth accrues from waste reduction, recy- cling, production of by-products, etc. (Adelaja 1992). Capital productivity growth was rapid during the 1964-84 period. Hence, capital intensity declined. Material intensity increased, however, suggesting that materials were substituted for capital (see Ta- ble 3). Production labor intensity increased while nonproduction labor intensity decreased (see Table 3). Hence, nonproduction labor productivity growth outpaced growth in production labor pro- ductivity. The apparent substitution of production for nonproduction labor is consistent with Oi’s ar- gument that less capital-intensive technologies re- quire less management and nonproduction work- ers, and more production workers. Apparently, as production became less capital-intensive, the rela- tive demand of New Jersey food processors for nonproduction labor, much of which is manage- Ad.daja Productivity Growth and Input Mix Changes in Food Processing 27 ment labor, declined. Overall, the changes in input mix in the sector were toward less nonproduction labor and capital intensities, but greater production labor and material intensities. The declining New Jersey shares of U.S. food- processing activities have been attributed to chang- ing transportation economics, increasing costs of acquiring raw material locally (due to declining local supply of farm products), stringent waste- disposal regulation, and high fixed costs of pro- duction (e.g., higher real estate costs) in New Jer- sey (Lopez and Henderson). Adelaja (1988) ar- gued that slower TFP growth in New Jersey food processing, relative to the rest of the U. S., also made New Jersey a less attractive location. Results of this study provide additional information on New Jersey’s food-processing industry. Specifi- cally, the results explain the input mix changes and productivity growth that accompanied the de- cline of New Jersey’s share of food-processing ac- tivities in the U.S. The Subsectors, 1964-84 Note that material productivity increased in all subsectors except the beverage group, which is highly material-intensive. The trend in beverage production in New Jersey has been from full pro- cessing to the mere dilution of concentrates shipped in from other states (Adelaja 1988). Hence, the significant increase in material inten- sity and the decline in material productivity in bev- erage production is not surprising. Consistent with aggregate-sector findings, ma- terial productivity growth was outpaced by growth in other inputs’ productivities in four of the eight subsectors (dairy, bakery, sugar and confection- ery, and beverage). Material intensity increased in these same subsectors. Material intensity also in- creased in the preserved fruit and vegetables, grain mill, and miscellaneous-products subsectors. Con- sequently, the only exception to increased material intensity is meat processing, where material pro- ductivity growth outpaced growth in productivities of other inputs. The relatively rapid growth in ma- terial productivity in the meat subsector may re- flect greater incentives to implement material and waste-reducing technologies due to the heavy reg- ulation of material waste from meat processing. Consistent with the pattern for the aggregate sector, capital intensity declined in all subsectors, but capital productivity increased. Growth in non- production workers’ productivity exceeded that of production workers’ productivity in most subsec- tors. The W values fiuther suggest that production labor was generally substituted for nonproduction labor. This is consistent with the finding for the aggregate sector. Contrary to the trend for the ag- gregate sector and most subsectors, production la- bor productivity actually declined in the meat (SIC 201) and grain mill (SIC 204) subsectors. The N’ values indicate that in both subsectors substitution of production labor for nonproduction labor was significant. The relative growth rates of TFP are worth not- ing. For example, TFP growth was most rapid in the bakery subsector (45% gain). Bakery was fol- lowed by sugar and confectionery (34% gain), pre- served fruit and vegetables (30Y0gain), dairy and beverage (24% gain), and miscellaneous products (22% gain). Grain mill products experienced the least gain in total factor productivity (3% gain). TFP gain in meat production was also limited (lo%). The Energy Crisis Given some of the recent events in the Middle East, there is growing concern among economists that drastic shocks in energy prices, similar to what happened in 1973, might again occur. Changes in intensity values and productivity indexes in 1973 should generally reflect potential impacts of future energy price hikes on productivity and the struc- ture of production. In the aggregate sector and all subsectors, total factor productivity declined dras- tically in 1973. Similarly, all partial factor produc- tivity indexes declined in 1973, except in the case of production labor productivity, which declined in grain mill production. It appears, therefore, that because they result in greater declines in output than in inputs, energy price shocks are usually pro- ductivity-dampening in the short run. The excep- tion in the case of grain mill production is difficult to explain. In the aggregate sector, material productivity fell more than did productivities of other inputs in 1973. Hence, material intensity increased, while the intensities of other inputs declined. Producers therefore seem less capable of reducing material consumption (compared with other inputs) when energy price shocks occur. This is an indication of the strong complementarily between materials and output. Apparently, recessions resulting from en- ergy price shocks would result in greater labor and capital unemployment than in material unemploy- ment. This implies that farmers are not as likely to get hurt as would suppliers of other resources to the food-processing sector when energy prices surge. In the aggregate sector, the energy crisis re- sulted in greater unemployment of nonproduction 28 April 1992 NJARE than production workers. This is not surprising considering that the former are more highly paid and that the energy crisis also reduced capital in- tensity. Greater unemployment of nonproduction than production workers is likely to accompany future increases in energy prices. TFP declined by 15% in 1973. Also, the 1973 decline in material productivity exceeded those of other inputs. Following the weighting procedure in (12), material inputs’ true contribution to the 1973 decline in TFP k estimated to be 64%. The impli- cation is that material productivity changes are very important, especially during periods of en- ergy price shocks. Now, examine the impacts of the energy crisis on total and partial factor productivities in the sub- sectors. In the meat, bakery, sugar and confection- ery, beverage, and miscellaneous-product groups, the impacts were similar to the aggregate case in that material productivity declined more than the productivities of other inputs. Also, consistent ‘with the aggregate sector case, the instantaneous effect of the energy price shock involved increased material intensity and material-capital substitution in most subsectors. In the cases of dairy, preserved fruit and vege- tables, and grain mill products, greater decline in nonproduction labor productivity than in the pro- duc~ivities of materials, capital, and production la- bor resulted from the energy crisis. Hence, man- agement workers in these subsectors seem to enjoy an unemployment buffer when energy prices rise. Note also from the intensity values in Table 3 that the energy price shock involved greater unemploy- ment of production than nonproduction labor in meat, dairy, preserved fruit and vegetables, and sugar and confectionery processing, while it re- sulted in greater unemployment of nonproduction than production labor in the rest of the subsectors. Conclusion This paper combines total and partial factor pro- ductivity indexes in an innovative way to analyze productivity growth and input mix changes in New Jersey’s food-processing industry. While it may have some limitations, the approach allows better accounting of contributions of specific inputs to total factor productivity growth. Results for the entire 1964-84 period suggest a 28% overall pro- ductivity growth and slower material productivity growth than labor and capital productivity growth. However, given the high cost share of material inputs, gains in the efficiency of use of materials explain almost half of the growth in total factor productivity. An implication of this is that material productivity growth, which is typically ignored in productivity studies, is an important component of productivity growth in food processing. Capital productivity grew rapidly in the sector. Simultaneously, the relative demand for nonpro- duction labor, vis-il-vis production labor, declined due to the complementarily between the former and capital-intensive technologies. Information obtained on input intensities in the sector are useful in analyzing the trends in input mix and in corre- lating these with the pattern of productivity growth. Such analysis is hardly ever conducted in conjunction with productivity analysis. An objective of this study was to examine the immediate impact of the 1973 energy crisis. Re- sults indicate that productivities of all inputs (as well as total factor productivity) tend to fall in the short run when energy price shocks occur. The decline in material productivity exceeds the de- clines in the productivities of other inputs, while the reduction in material use is less than reductions in other inputs. Hence, farmers supplying food processors are better protected than other resource suppliers when energy price shocks occur. The 1973 energy crisis also resulted in greater unem- ployment of nonproduction than production labor, suggesting that the former is more vulnerable in times of energy price increases. References Adelaja, A.O. The Agriculture and Food Complex of New Jer- sey. New Jersey Agricultural Experiment Station Publica- tion no. SR-02521-1-88. June 1988. Adelaja, A. O. “Material Productivity in Food Manufactur- ing. ” American Journal of Agricultural Economics 74, no. 1(1992):177–85. Azzam, A. M., and E. Pagoulatos. “Testing Oligopolistic and Oligopsonistic Behavior: An Application to the U.S. Meat-Packing Industry. ” Journal of Agricultural Econom- ics 41( 1990):260-70. Christensen, L. R., D. Cummings, and D. W. Jorgenson. c‘Eco- nomic Growth, 1947– 1973: An International Compari- son, ” In New Developments in Productivity Measurement and Analysis, edited by J. W. Kendrick and B .N. Vaccrrra. Chicago: National Bureau of Economic Research, 1980, Connor, J. M., R.T. Rogers, B.W. Marion, and W.F. Mueller. The Food ManufacnrringIndustries: Structure, Strategies, Performance and Policies. Lexington, MA: Lexington Books, 1988, Evenson, R. E., D. Landau, and D. Ballou. “Agricultural Pro- ductivity Measures for U.S. States, 1950-82. ” In Evak- ating Agricultural Research and Productivity. Miscella- neous Publications no. 52-1987. Minnesota Agricultural Experiment Station, University of Minnesota, 1987. Heien, D.M. “Productivity in U.S. Food Processing and Dis- Adelaja Productivity Growth and Input Mix Changes in Food Processing 29 tribution. ” American Journal of Agricultural Economics 65(1983):297-302. Jorgenson, D. W., F. Gollop, and B. Fraumeni. Producrivi~ and U.S. Economic Growth. Cambridge: Harvard Univer- sity Press, 1987. Lee, D.R. “Labor Market Dynamics in the Food Sector. ” American Journal of Agricultural Economics 70(1988): 90-102. Lopez, R. A., and N.R. Henderson. 6‘The Determinants of Lo- cation Choices for Food Processing Plants. ” Agribusiness 5(1989):619-32. Oi, W .Y. “Slack Capacity: Productive or Wasteful?” Ameri- can Economic Review 21( 198 1):64-69. Pratten, C. “A Survey of Economies of Scale. ” In Research on the “Cost of Non-Europe, ” Basic Findings. Luxembourg: Office for Official Publications of the European Commu- nities, 1988. Schroeter, J. R. “Estimating the Degree of Market Power in the Beef Packing Industry. ” Reviewof .Ecorromicsand Staris- rics 70(1988):158-62. Schroeter, J. R., and A. Azzam. “Measuring Market Power in Multi-Product Oligopolies: The U.S. Meat Industry. ” Ap- plied Economics 22(1990):1365-76, U.S. Department of Commerce. Bureau of the Census. Annual Survey of Manufacturers. Various years, —. Census of Manufacturers. Various years. U.S. Department of Labor. Bureau of Labor Statistics. Pro- ducer Price Indexes. Various years.